الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 140 |  |  |
| | | from | 140 | | |
Abstract
The importance of text processing is in a continuous increasing because text has become a vital part of all our daily interactions in e-books, articles, emails, and social media. The precision of Natural language Processing (NLP) applications is directly proportional to how far the machine understand the actual meaning of text. As a result of this necessity text Semantic Analysis (SA) has become an essential trend for computational linguistics and researchers in NLP. The generic objective of SA is to make machines automatically understand the accurate meaning of text. This thesis is considered a step forward towards a machine that understands and infers the meanings and relations in Arabic language texts which affects positively in the performance of different NLP applications. The first system is proposed to address the problem of semantic relatedness measurement in Arabic short texts by providing different approaches for text semantic representation enrichment. The results proved the superiority of the proposed system from the best previous approaches up to 99%. The second system is proposed to extract and infer Semantic Relations by means of Syntactic Dependency Relations (SDRs). The adjectival relations are used as a type of SDRs. The proposed system can map the adjectival relations onto valuable semantic relations such as finding synonyms and recognizing pronoun referees.